Files
git.stella-ops.org/docs/features/checked/jobengine/jobengine-audit-ledger.md

40 lines
4.1 KiB
Markdown

# Orchestrator Audit Ledger
## Module
Orchestrator
## Status
IMPLEMENTED
## Description
Append-only audit ledger tracking all orchestrator job lifecycle state changes, rate-limit decisions, and dead-letter events with tenant-scoped isolation.
## Implementation Details
- **Modules**: `src/JobEngine/StellaOps.JobEngine/StellaOps.JobEngine.Core/Domain/`, `src/JobEngine/StellaOps.JobEngine/StellaOps.JobEngine.Core/DeadLetter/`, `src/JobEngine/StellaOps.JobEngine/StellaOps.JobEngine.Infrastructure/Ledger/`, `src/JobEngine/StellaOps.JobEngine/StellaOps.JobEngine.WebService/`
- **Key Classes**:
- `AuditEntry` (`src/JobEngine/StellaOps.JobEngine/StellaOps.JobEngine.Core/Domain/AuditEntry.cs`) - audit entry model with action type, actor, tenant, timestamp, and metadata
- `RunLedger` (`src/JobEngine/StellaOps.JobEngine/StellaOps.JobEngine.Core/Domain/RunLedger.cs`) - run-level ledger tracking execution history
- `SignedManifest` (`src/JobEngine/StellaOps.JobEngine/StellaOps.JobEngine.Core/Domain/SignedManifest.cs`) - signed manifest for tamper-evident ledger export
- `LedgerExporter` (`src/JobEngine/StellaOps.JobEngine/StellaOps.JobEngine.Infrastructure/Ledger/LedgerExporter.cs`) - exports ledger data for compliance and audit
- `AuditEndpoints` (`src/JobEngine/StellaOps.JobEngine/StellaOps.JobEngine.WebService/Endpoints/AuditEndpoints.cs`) - REST API for querying audit ledger entries
- `LedgerEndpoints` (`src/JobEngine/StellaOps.JobEngine/StellaOps.JobEngine.WebService/Endpoints/LedgerEndpoints.cs`) - REST API for ledger export and querying
- `AuditLedgerContracts` (`src/JobEngine/StellaOps.JobEngine/StellaOps.JobEngine.WebService/Contracts/AuditLedgerContracts.cs`) - API contracts for audit responses
- `DeadLetterEntry` (`src/JobEngine/StellaOps.JobEngine/StellaOps.JobEngine.Core/Domain/DeadLetterEntry.cs`) - dead-letter entry in the audit trail
- `DeadLetterNotifier` (`src/JobEngine/StellaOps.JobEngine/StellaOps.JobEngine.Core/DeadLetter/DeadLetterNotifier.cs`) - notifies on dead-letter events
- `ErrorClassification` (`src/JobEngine/StellaOps.JobEngine/StellaOps.JobEngine.Core/DeadLetter/ErrorClassification.cs`) - classifies errors for dead-letter categorization
- `ReplayManager` (`src/JobEngine/StellaOps.JobEngine/StellaOps.JobEngine.Core/DeadLetter/ReplayManager.cs`) - manages replay of dead-letter entries
- `DeadLetterEndpoints` (`src/JobEngine/StellaOps.JobEngine/StellaOps.JobEngine.WebService/Endpoints/DeadLetterEndpoints.cs`) - REST API for dead-letter management
- `TenantResolver` (`src/JobEngine/StellaOps.JobEngine/StellaOps.JobEngine.WebService/Services/TenantResolver.cs`) - ensures tenant-scoped audit isolation
- **Interfaces**: `ILedgerExporter` (`src/JobEngine/StellaOps.JobEngine/StellaOps.JobEngine.Infrastructure/Ledger/ILedgerExporter.cs`), `IAuditRepository` (`src/JobEngine/StellaOps.JobEngine/StellaOps.JobEngine.Infrastructure/Repositories/IAuditRepository.cs`), `IDeadLetterRepository` (`src/JobEngine/StellaOps.JobEngine/StellaOps.JobEngine.Core/DeadLetter/IDeadLetterRepository.cs`), `ILedgerRepository` (`src/JobEngine/StellaOps.JobEngine/StellaOps.JobEngine.Infrastructure/Repositories/ILedgerRepository.cs`)
- **Source**: Feature matrix scan
## E2E Test Plan
- [ ] Trigger a job state transition and verify an `AuditEntry` is created in the ledger with action type, actor, and timestamp
- [ ] Query the audit ledger via `AuditEndpoints` with a time range filter and verify only matching entries are returned
- [ ] Verify tenant isolation via `TenantResolver`: create audit entries for two tenants and verify each tenant only sees their own entries
- [ ] Trigger a dead-letter event and verify it appears in both the `DeadLetterEntry` store and the audit ledger
- [ ] Export the audit ledger via `LedgerExporter` and verify the export contains all entries within the specified range
- [ ] Replay a dead-letter entry via `ReplayManager` and verify the replay action is also audited
- [ ] Verify `ErrorClassification` categorizes different error types correctly (transient, permanent, unknown)
- [ ] Query dead-letter entries via `DeadLetterEndpoints` and verify pagination and filtering work